Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Talos Regression :( a11y increase 41.5% on Linux x64 Firefox

0 views
Skip to first unread message

nob...@cruncher.build.sjc1.mozilla.com

unread,
Jan 20, 2011, 4:35:54 AM1/20/11
to dev-tree-...@lists.mozilla.org
Regression :( a11y increase 41.5% on Linux x64 Firefox
------------------------------------------------------
Previous: avg 282.800 stddev 1.887 of 30 runs up to revision efbf1fa4c70e
New : avg 400.300 stddev 4.563 of 5 runs since revision de195d1171d6
Change : +117.500 (41.5% / z=62.257)
Graph : http://mzl.la/hiILzT

Changeset range: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=efbf1fa4c70e&tochange=de195d1171d6

Changesets:
* http://hg.mozilla.org/mozilla-central/rev/d5e9eddf4acd
: Alexander Surkov <surkov.a...@gmail.com> - Bug 606924 - create tree on content insertion, r=davidb, fer, a=blocking2.0Final+
: http://bugzilla.mozilla.org/show_bug.cgi?id=606924

* http://hg.mozilla.org/mozilla-central/rev/de195d1171d6
: Alexander Surkov <surkov.a...@gmail.com> - Bug 625009 - text offsets don't get updated when text of first child text accessible is changed, r=davidb, a=blocking2.0Final+
: http://bugzilla.mozilla.org/show_bug.cgi?id=625009

Bugs:
* http://bugzilla.mozilla.org/show_bug.cgi?id=625009 - text offsets don't get updated when text of first child text accessible is changed
* http://bugzilla.mozilla.org/show_bug.cgi?id=606924 - create complete accessible tree

nob...@cruncher.build.sjc1.mozilla.com

unread,
Jan 20, 2011, 4:45:31 AM1/20/11
to dev-tree-...@lists.mozilla.org
Regression :( a11y increase 42.5% on Linux Firefox
--------------------------------------------------
Previous: avg 335.133 stddev 3.043 of 30 runs up to revision efbf1fa4c70e
New : avg 477.500 stddev 2.894 of 5 runs since revision de195d1171d6
Change : +142.367 (42.5% / z=46.791)
Graph : http://mzl.la/fGR478

nob...@cruncher.build.sjc1.mozilla.com

unread,
Jan 20, 2011, 7:55:20 AM1/20/11
to dev-tree-...@lists.mozilla.org
Regression :( a11y increase 59% on Win7 Firefox
-----------------------------------------------
Previous: avg 201.883 stddev 2.476 of 30 runs up to revision efbf1fa4c70e
New : avg 321.000 stddev 5.074 of 5 runs since revision de195d1171d6
Change : +119.117 (59% / z=48.101)
Graph : http://mzl.la/dJZRUl

nob...@cruncher.build.sjc1.mozilla.com

unread,
Jan 20, 2011, 7:55:21 AM1/20/11
to dev-tree-...@lists.mozilla.org
Regression :( a11y increase 62.6% on XP Firefox
-----------------------------------------------
Previous: avg 184.550 stddev 9.059 of 30 runs up to revision efbf1fa4c70e
New : avg 300.000 stddev 5.755 of 5 runs since revision de195d1171d6
Change : +115.450 (62.6% / z=12.745)
Graph : http://mzl.la/fypZ2M

Daniel Holbert

unread,
Jan 20, 2011, 2:55:17 PM1/20/11
to dev-tree-...@lists.mozilla.org, surkov.a...@gmail.com
So this morning we had these regression reports, all with the same
blamed push (below):
> Talos Regression :( a11y increase 41.5% on Linux x64 Firefox
> Talos Regression :( a11y increase 42.5% on Linux Firefox
> Talos Regression :( a11y increase 59% on Win7 Firefox
> Talos Regression :( a11y increase 62.6% on XP Firefox

Looks like a real regression... surkov, do you know if this was expected?

~Daniel


On 01/20/2011 01:35 AM, nob...@cruncher.build.sjc1.mozilla.com wrote:
> Regression :( a11y increase 41.5% on Linux x64 Firefox
> ------------------------------------------------------
> Previous: avg 282.800 stddev 1.887 of 30 runs up to revision efbf1fa4c70e
> New : avg 400.300 stddev 4.563 of 5 runs since revision de195d1171d6
> Change : +117.500 (41.5% / z=62.257)
> Graph : http://mzl.la/hiILzT
>

> Changeset range: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=efbf1fa4c70e&tochange=de195d1171d6
>
> Changesets:
> * http://hg.mozilla.org/mozilla-central/rev/d5e9eddf4acd
> : Alexander Surkov<surkov.a...@gmail.com> - Bug 606924 - create tree on content insertion, r=davidb, fer, a=blocking2.0Final+
> : http://bugzilla.mozilla.org/show_bug.cgi?id=606924
>
> * http://hg.mozilla.org/mozilla-central/rev/de195d1171d6
> : Alexander Surkov<surkov.a...@gmail.com> - Bug 625009 - text offsets don't get updated when text of first child text accessible is changed, r=davidb, a=blocking2.0Final+
> : http://bugzilla.mozilla.org/show_bug.cgi?id=625009
>
> Bugs:
> * http://bugzilla.mozilla.org/show_bug.cgi?id=625009 - text offsets don't get updated when text of first child text accessible is changed
> * http://bugzilla.mozilla.org/show_bug.cgi?id=606924 - create complete accessible tree

> _______________________________________________
> dev-tree-management mailing list
> dev-tree-...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-tree-management

David Bolter

unread,
Jan 20, 2011, 3:31:32 PM1/20/11
to Daniel Holbert, dev-tree-...@lists.mozilla.org, surkov.a...@gmail.com
I didn't expect it would be this bad. I'll discuss with Alexander when
he wakes up.

Cheers,
David

Ehsan Akhgari

unread,
Jan 20, 2011, 8:24:30 PM1/20/11
to David Bolter, Daniel Holbert, dev-tree-...@lists.mozilla.org, surkov.a...@gmail.com
What were the results of the discussion?

david bolter

unread,
Jan 20, 2011, 8:37:47 PM1/20/11
to Ehsan Akhgari, Daniel Holbert, dev-tree-...@lists.mozilla.org, surkov.a...@gmail.com
One of us will post to the list once we've had the discussion (or before).
;)

D

David Bolter

unread,
Jan 21, 2011, 3:01:38 PM1/21/11
to Ehsan Akhgari, Daniel Holbert, dev-tree-...@lists.mozilla.org, surkov.a...@gmail.com
Sorry for the delay as I haven't made contact with Alexander yet. I
think we may back this out but we're currently testing and profiling to
get more data.

Cheers,
David

Alexander Surkov

unread,
Jan 23, 2011, 3:11:43 AM1/23/11
to Daniel Holbert, David Bolter, dev-tree-...@lists.mozilla.org
Hi.

The first patch of bug 606924 "create tree on content insertion"
regressed us. I expected possible regressions though I didn't spot so
bad results in tests I've done. The same time I missed that I need to
check talos tests.

The next patches in that bug improves results at 10% in talos tests on
my local machine. I still analyze trying to figure out how to improve
the perf.

Thank you.
Alex.

0 new messages